GIS 5935 - Lab 4 - Building Networks

The first part of this week's assignment was to run through some ArcGIS Network Analyst tutorial exercises.  In the second part of the lab we practiced building network datasets.  Using a provided Streets feature class we created a new network dataset.  The first network dataset modeled turns but did not have restricted turns and did not model traffic or driving directions.  The network was then built and added to an ArcMap data frame.  Using the Network Analyst toolbar we started a new Route Analysis.  Stops along the route were created by uploading a provided Facilities point layer.  The route was solved and produced a path that took 105.49 total minutes to travel and was 10.06 kilometers in length.

Next we returned to the Properties window of the new network dataset and under the Turns tab added RestrictedTurns.   RestrictedTurns was a provided feature class that specified conditions for when turns were and were not allowed.  This feature class also specified whether the restriction applied to  specific transportation modes; such as cars, trucks, pedestrians, etc.  The network dataset had to be rebuilt after RestrictedTurns was added.  The route analysis was rerun.  Despite adding RestrictedTurns the modeled route was unaffected by the turn restrictions and resulted in the same travel time and path length as the first route analysis.

For our final analysis we had to start over and use a copy of the supplied geodatabase and create a brand new network dataset.   This time we used the same settings as the previous network setup but we chose to model traffic using two provided profile tables, Patterns and Streets_Patterns.  These tables linked street segment features with traffic speed profiles for each day of the week.  The new network dataset was added to a data frame and another route analysis was performed using the same Facilities points as stops along the route.  Even with historical traffic data added to the network dataset there was no change in the results of the route analysis.   The total travel time and path length remained the same 105.49 minutes and 10.06 km.

I have a new appreciation for the level of detail that is needed in a Streets layer to create a more accurate network dataset.  The Network Analyst can provide a user with many options in analyzing potential routes and their associated costs.

GIS5935 - Lab 3 - Determining Quality of Road Networks - Completeness

In Lab 3, the second part of our analysis of road network quality, we focused on completeness of the networks.  Our task was to provide a comparison between street centerline roads and TIGER 2000 roads for Jackson County, Oregon.  We were provided a county boundary layer as well as a grid of 5km by 5km polygons to overlay the roads for analysis.  A simple method of measuring completeness assumes more length means more complete.  We reported the total length for each road network for the entire county.   The TIGER network contained more total length than the centerlines and therefore was more complete.  However, our assigned reading by P.A. Zandbergen, et. al. explains the brief history and the positional accuracy issues with the TIGER 2000 dataset.

Determining the completeness for each polygon grid was obtained by breaking up the road segments using the polygon boundaries and determining the length of each road segment contained inside each respective boundary.  Using the Summarize option within the attribute table I was able to create a table containing the segment count for each grid and the total length of all the segments within each grid.   By subtracting the TIGER road length sum from the Centerline road length sum for each grid I was able to determine which road segment was more complete per grid.  The TIGER network was more complete than the Centerline network in 162 out of 297 grids.   The Centerline network was more complete than the TIGER network in 134 of the 297 grids.  One grid polygon did not contain any road segments.  After calculating the % difference in network length, I was able to create this choropleth map depicting road network completeness.  Difference values that were positive percentages meant that the Centerline network was more complete.  Polygons with negative percentages meant that the TIGER network was more complete.  Using completeness in conjunction with the positional accuracy analysis we performed in last week's lab aids the end user in determining the quality of the road network.

T

Completeness of Road Networks - A Comparison Between TIGER Roads and Jackson County, Oregon Centerlines

GIS5935 - Lab 2 - Determining Quality of Road Networks

Our task this week was to evaluate the quality of two different road networks for Albuquerque, New Mexico.  The first step was to create "New Network Datasets" for each of the provided street shapefiles.  This generated "junction" point shapefiles for each street network.  

Next, using examples of good intersections I identified twenty "test" points using the ABQ junctions.  The overall criteria was to be:

1. a "good" intersection 

2. Meet the sampling rules (min of 20% in each quadrant and > 10% of diameter apart)

3. Matching locations in both street datasets

I used the draw tool to "eyeball" the quadrants and diameter of the "study area" so my test points would meet the 2nd criteria.

I used the ortho images provided to create a new "reference" shapefile and digitized reference points using the twenty test locations from the ABQ junction dataset.

In order to compare each street dataset with my reference set I added a short integer field to all three shapefiles (reference, ABQ_Streets, StreetMapsUSA).  It took a long time to edit the attribute table of each shapefile.  

Using the AddXY Coordinates tool, under Data Management Tools > Features, I was able to quickly add the X and Y coordinate values for each point in the three shapefiles.

I used the table to Excel tool to convert the attribute tables of my three shapefiles to Excel.  I combined all the Excel worksheets into one workbook and used the provided NSSDA spreadsheet to calculate the necessary values.  The following results were reported.

StreetMapUSA Road Network

Horizontal Positional Accuracy           Tested 367.260 feet horizontal accuracy at 95% confidence level

ABQ_Streets Road Network

Horizontal Positional Accuracy           Tested 18.810 feet horizontal accuracy at 95% confidence level

This was a very time consuming lab assignment for me but I realized the importance of data quality.

Twenty Reference/Test Points Used to Determine Horizontal Accuracy of Two Different Road Networks

GIS 5100 - Final Project - Spatial Access to AZA Accredited Zoos in Metro Chicago

The metro Chicago area is extremely fortunate to be geographically located near two distinct zoos.  The Brookfield Zoo in Brookfield, Illinois and the Lincoln Park Zoo in downtown Chicago are both accredited by the Association of Zoos and Aquariums (AZA).  The Brookfield Zoo is set within a 216-acre nature park and is home to over 2,000 animals and is also an accredited arboretum.  The Lincoln Park Zoo is a free admission, 35-acre zoo residing within its namesake’s 1,208-acre lakefront park along Chicago’s north side and his home to over 1,000 animals.  I wanted to perform a spatial accessibility analysis of this population’s access to these two accredited zoos.  As noted by a 2016 article in The Hill, an American political journalism newspaper, “zoos, aquariums and conservation parks are essential if we want to preserve the magnificent creatures with whom we share the Earth” (Ganzert, 6/14/16).  The information provided by an accessibility study could provide valuable information to these two zoos for marketing and outreach programs.  To assess the spatial accessibility to these zoos I used 2010 US Census tract and population data for Cook and DuPage counties.

My analysis found that over two-thirds of the populations of Cook and DuPage counties have access to two distinct, accredited zoos within 15-20 miles of the centers of 2010 census tracts.  With such access it is hopeful that the residents of these counties will visit these “agents of conservation” as described by Rabb & Saunders guest essay in the Zoological Society of London’s International Zoo Yearbook over a decade ago (2005).  The goal is for experiences at these venues to inspire and motivate humans to value animals, creatures and natural resources and take action to ensure their preservation for the benefit of us all. 

One of the limitations of this analysis is the use of the centroid of each census tract to calculate the distance to each zoo.  This is not an exact measurement of distance from each residence nor does it take into account travel time or transportation method.

For future work regarding this analysis I would consider the following: five of the US Census tracts in the chosen study area have 2010 population counts of zero and two of those have acreage over water.  In retrospect I should have removed these tracts before analysis.  If time permitted I would have continued statistical analysis and created CDF scatterplots for other demographic data, such as race/ethnicity and poverty levels.  I would also expand the study area to include all counties considered as “Chicagoland” by the Chicagoland Chamber of Commerce (Cook, DuPage, Kane, Lane, McHenry, and Will).  Additionally, I would add the John G. Shedd Aquarium to the Zoo feature class as it, too is an accredited member of the AZA and provides another venue for the residents of metro Chicago to encounter and appreciate wildlife.

While I am deeply saddened about the recent passing of my uncle, Dr. George B. Rabb, President Emeritus of the Chicago Zoological Society, I am truly honored to have gotten to know him and his life’s work more intimately through this final project.

Study Area of GIS 5100 Final Project on Spatial Access to AZA Accredited Zoos in Metro Chicago

Spatial Access to Brookfield Zoo from 2010

Census Tracts in Cook and DuPage Counties

Cumulative Distribution Function of Spatial Access to Brookfield Zoo

Spatial Access to Lincoln Park Zoo from 2010

Census Tracts in Cook and DuPage Counties

Cumulative Distribution Function of Spatial Access to Lincoln Park Zoo